Depletion of ID3 enhances mesenchymal stem cells therapy by targeting BMP4 in Sjögren’s syndrome

Mesenchymal stem cell (MSCs) transplantation has been used to treat Sjögren’s syndrome (SS) based on the immunoregulatory properties of MSCs. However, the effectiveness need improving and its underlying intrinsic mechanisms remain largely unknown. Here, we show that Id3 is upregulated in bone marrow-derived MSCs (BMMSCs) isolated from NOD/ShiLtJ mice, a widely used SS model, compared with ICR mice as control, suggesting that it functions in SS development and therapy. Transplantation of Id3-deficient BMMSCs rescues salivary gland function more effective than wild-type BMMSCs in NOD/ShiLtJ mice. Mechanistically, we show that ID3 negatively regulated BMP4 expression by preventing binding of basic helix–loop–helix protein E2A to the promoter of the Bmp4 gene. BMP4 in turn promoted PGE2 production in MSCs, and exhibited enhanced suppressive activities of T-cell proliferation and Th1 differentiation. Importantly, BMMSCs from SS patients showed significantly lower BMP4 and PGE2 expression than those from healthy individuals. Taken together, our findings revealed the targeting Id3 may be therapeutically useful for improving MSC immunoregulation and effectiveness of MSCs therapy for SS.


Introduction
Sjögren's syndrome (SS) is a chronic autoimmune disorder that typically impairs the function of exocrine glands, mainly the salivary and lacrimal glands, resulting in symptoms of dry mouth and eyes 1 . Dysfunction of immune systems have been shown to play a significant role in SS pathogenesis, including innate immune system, B cell activation, and resultant activation of T cellsmainly type 1 Thelper (Th1) cells. Current treatments for SS patients include steroids and nonsteroidal antiinflammatory drugs, which relive symptoms, but do not cure the disease and have serious side effects with longterm use. Mesenchymal stem cells (MSCs) have immunoregulatory properties and show therapeutic effects on experimental and clinical autoimmune diseases and inflammation, including SS, Crohn's disease, systemic lupus erythematosus, and rheumatoid arthritis [2][3][4][5] . However, the clinical trial showed the efficacy MSC-based therapies are not consistent, partial patients present useless [6][7][8] . Thus, uncovering the mechanisms that maintain MSCs immunoregulatory properties or MSC-based therapeutic effects is important for improving MSC-based therapies.
Inhibitor of DNA binding proteins (IDs), including ID1-4, are helix-loop-helix (HLH) transcription factors that lack the basic region required for DNA binding found in other HLH proteins 9 . IDs bind to basic (b) HLH E proteins and disrupt the formation of dimers, thus inhibiting DNA binding and transcriptional activity. E proteins have been found to play key roles in the differentiation and biological properties of various cells. Subsequent studies have uncovered E proteins inhibit cell cycle progression and implicated in cellular senescence activate through activate transcription of the CDK inhibitors 10 . As key regulators of E proteins, ID proteins have been implicated in the proliferation and self-renewal of embryonic stem cells, somatic stem cells, progenitors and hematopoietic stem cells [11][12][13][14] .
ID3 contributes to cellular differentiation and proliferation. It is involved in T-cell receptor (TCR)-mediated T-cell development 15,16 and in the differentiation of regulatory T cells (Tregs), Th17 cells, Th9 cells, and B cells [17][18][19] . Id3 −/− mice develop many symptoms similar to those found in Sjögren's syndrome, like immune cells chronically attack the lachrymal and salivary glands, resulting in impaired tear and saliva secretion. T-cell dysfunction has a prominent role in the development of SS in mice that lack Id3 (Id3 −/− ) 20,21 . However, the roles of ID3 in mediating the immunoregulatory function of MSCs in the development of SS is yet unknown.
We found that Id3 was upregulated in Bone marrowderived MSCs (BMMSCs) derived from NOD/ShiLtJ mice, which is SS model, when compared with ICR mice. Depletion of Id3 (Id3 −/− ) enhanced the inhibiting function of BMMSCs on T-cell proliferation and IFN-gamma production in vitro and in vivo. In this study, we showed that ID3 controls the immunosuppressive functions bone marrow MSCs by inhibiting BMP4 expression, which decreasing PGE2 production. Our findings provide novel insights into the immunoregulatory functions of MSCs in chronic inflammatory or autoimmune diseases.

Deletion of Id3 enhanced immunosuppression of BMMSCs to T-cell proliferation and IFN-γ production in vitro and in vivo
Firstly, we compared gene expression in BMMSCs between NOD/ShiLtJ mice, which is widely used SS model, and ICR mice as control. Among the reported SS related genes 22 , Id3 is the highest expressed gene in NOD/ShiLtJ mice ( Supplementary Fig. 1). We generate Id3 knockout (Id3 −/− ) mice and found BMMSCs from wild-type (WT) and Id3 −/− mice exhibited no significant differences in proliferation and apoptosis (Supplementary Fig. 2B, C). To identify the role of Id3 on immunoregulation function of MSCs, normal splenic CD4 + CD25 − naive T cells were cocultured with WT or Id3 −/− BMMSCs in the presence of anti-CD3 and anti-CD28 antibodies. T cells cultured with Id3 −/− BMMSCs showed much lower proliferative responses, as determined by carboxyfluorescein succinimidyl ester-dilution assays, than T cells cultured with WT BMMSCs (Fig. 1a, and Supplementary Fig. 2D, E). No significant differences were found in T-cell apoptosis between WT and Id3 −/− BMMSCs cultures ( Supplementary Fig. 3A, B). Moreover, Id3 −/− BMMSCs treatment decreased the frequency of interferon (IFN)-γ producing Th1 cells compared with WT BMMSCs treatment (Fig. 1b), but did not change the frequency of interleukin (IL)-4 producing Th2 cells and Foxp3 + Tregs (Supplementary Fig.  3C-F). These results indicated that the immunosuppressive function of BMMSCs was enhanced in the absence of Id3 in vitro.
Next, WT or Id3 −/− BMMSCs and CD4 + T cells (KJ-126 + ) derived from DO11.10 TCR-transgenic mice were mixed and injected into syngeneic BALB/cJ mice, who were then challenged by ovalbumin protein (peptide) administered by intravenous injection. Seven days later, flow cytometric analysis showed that there were fewer CD4 + KJ-126 + T cells in Id3 −/− BMMSC-injected BALB/cJ mice than in WT BMMSC-injected mice (Fig. 1c), and fewer Th1 cells within the injected CD4 + T-cell population (Fig. 1d). However, there were no changes in the number of Th2 cells or Tregs between these two groups ( Supplementary Fig. 3G-K). Thus, Id3 −/− BMMSCs also inhibited T-cell growth and Th1 cell differentiation to specific antigen stimulation in vivo. These results indicated that the immunosuppressive function of Id3 −/− BMMSCs was enhanced in vitro and in vivo.
(see figure on previous page) Fig. 1 Id3 −/− BMMSCs enhanced immunosuppression of Id3-deficient BMMSCs to T-cell proliferation and IFN-gamma production in vitro and in vivo. a CFSE analysis of cell proliferation for T cells cultured with bone marrow-derived mesenchymal stem cells (BMMSCs) form WT mice and Id3 −/− mice (WT and Id3 KO). b Frequencies of IFN-γ producing Th1 cell subsets in the presence of Id3 −/− BMMSCs or WT BMMSCs; WT or Id3 −/− BMMSCs and CD4 + T cells derived from DO11.10 TCR-transgenic mice were mixed at a ratio of 1:2 and injected into BALB/cJ mice. c, d Flow cytometric analyses showing changes in CD4 + T and Th1 cell number in BALB/cJ mice injected with CD4 + T cells from DO11.10 TCR-transgenic mice combined with Id3 −/− or WT BMMSCs and a subsequent ovalbumin challenge, compared with T cells only. e T-cell proliferation when cocultured with WT or Id3 −/− BMMSCs in Transwell plates. f-j Changes in PGE 2 , TNF-α, IL-10, IL-6 and inducible nitric oxide synthase (iNOS) expression in supernatants of WT or Id3 −/− BMMSCs, T cells and cocultures. k Effect of COX2 IL-10, TNF-α or IL-10 inhibition on immunosuppressive effects of WT and Id3 −/− BMMSCs. l PGE 2 secretion in response to COX2 inhibition. m ID3 expression, as assessed by Western blot, in human BMMSCs transfected with ID3-specific siRNA or siRNA only (Ctrl). n PGE 2 secretion, as assessed by ELISA, in control and ID3-knockdown human BMMSCs. o CFSE analyses of cell proliferation in T cells cultured with ID3-knockdown human BMMSCs versus control BMMSCs. Values are means ± SD (n = 5, three independent experiments). Student's t tests and one-way ANOVA. *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.001.
To elucidate whether Id3 −/− BMMSC-mediated immunosuppressive effects on T cells requires direct cell contact, we cultured normal CD4 + T cells with BMMSCs in a Transwell plate. T cells cultured with Id3 −/− BMMSCs showed much lower proliferative responses than T cells cultured with WT BMMSCs (Fig. 1e), suggesting that the inhibitory function of Id3 −/− BMMSCs occurs through secretion of cytokines or other soluble factors. Accordingly, we examined the expression of a panel of MSC-related immunoregulatory mediator and found that prostaglandin E 2 (PGE 2 ), IL-10, and tumor necrosis factor (TNF)-α were upregulated in the supernatants of Id3 −/− BMMSCs and T-cell cocultures, whereas IL-6 was downregulated ( Fig. 1f-j). Among the upregulated immunoregulatory mediator, PGE 2 showed the most significant increase (10-fold) in Id3 −/− BMMSC and T-cell cocultures. We attributed this increased PGE 2 expression to the Id3 −/− BMMSCs, because Id3 −/− BMMSCs monoculture (without T cells) also produced significantly higher levels of PGE 2 than WT BMMSCs (Fig. 1f). In contrast, TNF-α and IL-10 levels did not differ between Id3 −/− and WT BMMSCs monocultures (Fig. 1g, h). Importantly, inhibition of the activity of PGE 2 with celecoxib, a cyclooxygenase 2 (COX2) inhibitor, markedly decreased the inhibitory effect of Id3 −/− BMMSCs on Tcell proliferation; but that was not the case when TNF-α inhibitor pomalidomide 22,23 or an anti-IL-10 antibody was used (Fig. 1k, l, and Supplementary Fig. 3L). These results indicated that the enhanced immunosuppressive functions of Id3 −/− BMMSCs can be mainly attributed to PGE 2 . The effects of ID3 on immunosuppression in MSCs were also verified in human BMMSCs from healthy volunteers. Knockdown of ID3 in human BMMSCs by ID3-specific small interfering RNA (siRNA; Fig. 1m) resulted in higher levels of PGE 2 compared with control BMMSCs (Fig. 1n), and suppression of T-cell proliferation (Fig. 1o).

BMP4 mediate BMMSCs therapy for SS
In BMMSCs from NOD/ShiLtJ mice, BMP4 expression was significantly decreased compared with those from ICR mice (control mice) at both mRNA and protein levels ( Fig. 4a, b). PGE 2 levels were decreased in BMMSCs and submandibular glands from NOD/ShiLtJ mice compared with those from ICR mice (Fig. 4c, d). BMMSCs transplantation increases the salivary flow rate and reduced the area of inflammation in the submandibular glands, and decreased IFN-γ production T (Th1) cells in the submandibular glands. The therapeutic effects of allogeneic BMMSCs transplantation on NOD/ShiLtJ mice were abolished by inhibition of BMP4 with a specific inhibitor before the WT BMMSCs were transplanted (Fig. 4e-h).
Id3 −/− BMMSCs suppress SS symptom more efficiently than WT BMMSCs in NOD/ShiLtJ mice We next detected the immunoregulatory function of Id3 −/− BMMSCs in NOD/ShiLtJ mice. We injected WT, Id3 −/− BMMSCs, or Id3 −/− BMMSCs that were pretreated with a BMP4 or COX2 inhibitor, into NOD/ShiLtJ mice at 16 weeks of age. No differences in body weights or submandibular gland weights between all NOD/ShiLtJ mouse groups ( Supplementary Fig. 5A-C). WT BMMSCs-treated group exhibited improved salivary flow rate, reduced the area of inflammation and numbers of infiltrated CD4 + T cells and Th1 cells compared with Fig. 4 Effects of BMP4 on the therapeutic efficacy of BMMSCs in SS. a, b BMP4 mRNA and protein levels in bone marrow-derived mesenchymal stem cells (BMMSCs) derived from NOD/ShiLtJ and ICR mice. c, d Prostaglandin E 2 (PGE 2 ) levels in supernatant of BMMSCs and submandibular glands from NOD/ShiLtJ and ICR mice. e The saliva flow rate of WT mice BMMSCs-treated mice was significantly blocked with BMP4 inhibition. f, g H&E staining showed that the area of infiltrating in the submandibular glands of NOD/ShiLtJ mice following BMMSCs injection was significantly smaller than untreated NOD/Ltj mice, and this protective effect was significantly decreased with BMP4 inhibition. Yellow dotted and arrow showed area represent infiltrating area in the submandibular glands. h IFN-γ + T cells in the submandibular glands from BMP4 BMMSCs injection inhibition were upregulated compared with BMMSCs injection group (n = 5, three independent experiments). Values are means ± SD. Student's t tests and one-way ANOVA. *P ≤ 0.05; **P ≤ 0.01, scale bar: 200 μm. untreated NOD/ShiLtJ mice 2 weeks after BMMSCs infusion ( Fig. 5a-e). However, the salivary flow rate of Id3 −/− BMMSCs-treated mice was significantly higher than that of WT BMMSCs-treated mice (Fig. 5a). Consistent with the salivary flow rate improvement, Id3 −/− BMMSCs transplantation significantly reduced the area of inflammation in the submandibular glands compared with that in WT BMMSCs-treated NOD/ShiLtJ mice (Fig. 5b, c). Importantly, the improved therapeutic effects of Id3 −/− BMMSCs transplantation were completely abolished by pre-inhibition of BMP4 or COX2 with specific inhibitors (Fig. 5a-c). Moreover, the numbers of infiltrated CD4 + T cells and Th1 cells in the submandibular glands were decreased in the Id3 −/− BMMSCs-treated group compared with those in the WT BMMSCs-treated group, and these effects were blocked by pretreatment of Id3 −/− BMMSCs with a BMP4 or COX2 inhibitor (Fig. 5d, e). These data confirmed the biological significance of Id3 −/− BMMSCs-mediated immunosuppression of inflammation by upregulating BMP4 in an experimental SS model. Extending our findings in mice to humans, five pSS patients and five healthy volunteers were enrolled for bone marrow collection, after which their BMMSCs were cultured. BMP4 expression and PGE 2 secretion were significantly decreased in BMMSCs from pSS patients compared with those from healthy volunteers (Fig. 6a-c).

Discussion
The expansion of T cells plays an important role in many immune-related diseases, which can contribute to abnormal T-cell accumulation and chronic inflammation. Th1 cells produce high levels of IFN-γ, and are responsible for both phagocyte activation and production of opsonizing and complement-fixing antibodies 26 . They also contribute to the development of organ-specific autoimmune diseases, including SS, experimental autoimmune encephalomyelitis, collagen-induced arthritis, and inflammatory bowel disease 27 . In the present study, we showed that Id3 −/− BMMSCs suppressed T-cell proliferation and IFN-γ Th1 cells, which provided a novel cellular pathway to target in the treatment of immunerelated diseases. Our finding that Id3 −/− BMMSCs  21 . The enhanced immunoregulatory role of BMMSCs in Id3 −/− mice may have been masked by the dominant immunopathogenic role of T cells in these mice. In this study, we also found Id3 levels of BMMSCs in NOD/ShiLtJ mice were high expressed. As our results showed Id3 depletion enhance the immunoregulatory role of BMMSCs, combining with our previous study have showed the impaired immunoregulatory role of BMMSCs in NOD/ShiLtJ mice, the increased Id3 levels of BMMSCs would be a pathogenic factor induced a Sjögren's syndrome. This is different to Id3 knockout mouse, as there is different immunopathology between NOD/ShiLtJ mice and Id3 knockout mice. In Id3 knockout mouse model, there is no obvious change of Th1 cell related cytokine IFN-γ, IL-2, and TNFα, but have uncontrolled IL-4 production, which is the Th2 cell related cytokine 21 . This indicated that the Sjögren's syndrome in Id3 knockout mouse was majorly involved by Th2 cells. However, the T cell involved in NOD/ShiLtJ mice was dominated by Th1 cell 1 . The different mechanism in the two models may lead to the different performance of immunoregulatory role of BMMSCs. Thus, our finding that ID3 exhibits a negative role in immunoregulation by MSCs expands our knowledge of this key molecule in regulating the complex network between immune cells and nonimmune cells, such as MSCs.
ID3 contrasting role in T cells and MSCs in the present study is similar to TWIST1 (another basic helix-loop helix protein). TWIST1 genes encode basic helix-loophelix transcription factors involved in the expression of cytokine genes in inflammation 28 . TWIST1 regulates Th1-mediated acute and chronic inflammation. T cells isolated from chronically inflamed tissue with ulcerative colitis (UC) or Crohn's disease (CD) and rheumatoid arthritis expressed high levels of TWIST1 29 . Mice deficient for TWIST1 have severe systemic inflammation, demonstrating the central role of TWIST1 proteins in the regulation of inflammation. However, depletion of TWIST1 significantly increased the secretion of PGE2, IDO1 and decreased IL-6 secretion in MSCs, and has enhanced suppressive function on proliferation of CD3 + T cells and expression of IFN-γ Th1. It also mediated IFN-γ regulated the anti-inflammatory and immunomodulatory activities of MSCs 30 . The contrasted function of basic helix-loop-helix protein in MSCs and T cells demonstrating that its tissue specific function character.
PGE2 is the important soluble factors involved in exert their immunosuppressive effects of MSCs 31 . Our data showed that increased PGE2 was regulated by a novel ID3/BMP4/PGE 2 axis in MSCs. As the expression of ID3 is reportedly regulated by BMP 32 , our findings first time provide important insights into the mechanisms through which ID3 regulates BMP4 via prevention of E2A binding to the BMP4 promoter. Thus, our results revealed ID3 and BMP4 regulate each other's expression. As a member of the transforming growth factor superfamily, BMP has important roles in immunity and in the development of autoimmune diseases 33 . For example, BMP4 signaling is involved in the differentiation and activation of natural killer cells 34 . BMP4 and BMP5 are downregulated in osteoarthritis, rheumatoid arthritis, and systemic lupus erythematous 35,36 . The expression of ID3 protein has been reported to be regulated by BMP. In this study, we showed that BMP4 plays an important role in immunoregulation of MSC and MSC-based therapy for experimental SS through induction of PGE 2 . This reveals a mutual regulation between ID3 and BMP4, this would be the molecular target by which host interact with MSCs immunoregulatory properties or MSC-based therapeutic effects. PGE 2 is an important immune regulator 37,38 , and the decreased concentration of PGE 2 is associated with the risk of developing an autoimmune disease, such as multiple sclerosis, Crohn's disease, and multiple ulcerative colitis [39][40][41][42] . PGE 2 can exert pro-inflammatory and antiinflammatory effects depending on the context and target. The effects of PGE 2 on CD4 + cells vary depending on the CD4 + T-cell subset, PGE 2 concentration, and activation status of the cells 43 . Evidence has shown that PGE 2 inhibits the proliferation of T cells and suppresses IFN-γ production from Th1 cells 44 . In addition, our previous work has demonstrated that periodontal ligament MSCs suppress T-cell proliferation via PGE 2 45 . In our present study, we also showed PGE 2 function in the SS therapy, these should have implications for treatment option for SS patients.
Although ID3 expression is not impaired in salivary glandular epithelial cells, peripheral T cells, or labial salivary glands in these patients 46 . Our data confirmed the biological significance decreased of BMP4 and PGE 2 by MSCs in SS, demonstrating that BMP4/PGE 2 can be the cellular pathway to target in the treatment of SS in human patients. These findings provide novel insights into the intrinsic mechanisms of MSC immunoregulation and may facilitate the development of novel therapies for SS and other immune diseases. Mice C57BL/6 mice (female, 6-8 weeks old), BALB/cJ mice (female, 6-8 weeks old), and NOD/ShiLtJ mice (female, 16 weeks old) were obtained from Jackson Laboratories. E2a f/f Heb f/f ER-Cre + mice (on C57BL/6 mice background, female, 6-8 weeks old) and DO11.10 TCR-transgenic mice (on a BALB/cJ background, female, 6-8 weeks old) were purchased from Taconic (New York, USA). Id3 −/− mice (on C57BL/6 mice background) were bred under specific-pathogen-free conditions in the animal facility of the Capital Medical University.

pSS patients
All patients had to meet the classification criteria based on the revised American-European criteria (2002) for primary Sjögren's syndrome (pSS). In brief, these criteria comprise subjective criteria (ocular and oral symptoms) and objective criteria (ocular signs, histopathological signs [focus score ≥ 1] or positive parotid sialography, oral signs, and serological signs [presence of antinuclear antibodies, anti-SSA or anti-SSB antibodies]). pSS is diagnosed when 4 of the 6 aforementioned criteria are present, as long as histopathology or serology is positive, or when 3 of any 4 objective criteria are present. In addition, patients had to be informed of the investigational nature of this study. Exclusion criteria were (1) active, uncontrolled infections, and (2) end-stage organ failure.

BMMSCs culture
Mouse bone marrow-derived mesenchymal stem cells (BMMSCs) were prepared from the bone marrow of femurs and tibias harvested from 6-to 8-week-old mice. Human BMMSCs were obtained from the iliac bone marrow. All BMMSCs cultured in Dulbecco's modified Eagle's medium with low glucose (DMEM-LG) containing 10% fetal bovine serum (FBS, Gibco). The BMMSCs were subsequently incubated at 37°C in humid air with 5% CO 2 and 20% O 2 . The medium was replaced every 3 days after the initial plating, and then replated at a dilution of 1:3 and third passage was used for experimentation. The stem cell properties of BMMSCs were characterized using cell surface markers (Sca-1, CD34, CD44, and CD45) and analyzed by flow cytometry. The pure rate of MSCs is almost 95% (Supplementary Fig. 2A).

T-cell proliferation assay
Mouse naive (CD4 + CD25 − ) T cells were resuspended in rewarmed phosphate-buffered saline (PBS) at a final concentration of 5 × 10 6 cells/mL then labeled with 5 mM carboxyfluorescein succinimidyl ester (Invitrogen) according to the manufacturer's instructions. T cells were cocultured with BMMSCs from wild-type (WT) or Id3 −/− mice at a ratio of 1:2 for 4 days. Lymphocytes were then harvest and assayed by flow cytometry. The proliferation index was calculated using ModFit LT 3.0 software (Verity Software House, USA) as the average number of cell divisions.

Overexpressing plasmid construction and viral infection
Mouse full-length Id3 and E2A cDNA were constructed with standard methods and fused to a Hemagglutinin (HA) tag and M2-FLAG tag respectively, was produced. This sequence (HA-Id3 and Flag-E2A) was subcloned into the pQCXIN retroviral vector with AgeI and BamH1 restriction sites. For viral infections, BMMSCs form WT mice were plated overnight, then infected with retroviruses in the presence of polybrene (6 µg/mL, Sigma-Aldrich, St. Louis, MO, USA) for 6 h. After 48 h, infected BMMSCs were selected with different antibiotics. Ectopic Id3 and E2A overexpression in transduced BMMSCs were confirmed by Western Blot analysis.

Co-immunoprecipitation (Co-IP) assay
The assay was detected by HA-Tag Co-IP kit (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer's protocol. WT mice BMMSCs transfected with vector plasmid or HA-Id3 plasmid were washed carefully with pre-chilled PBS two times; then, cold RIPA lysis buffer was added and cell lysate was collected. Ten microliters of anti-HA agarose slurry was added into each labeled spin column with gentle end-over-end mixing at 4°C for overnight. Then wash with TBS-T and Elution Buffer (10 μL) was added to the anti-c-HA agarose. At last the sample was heated for 10 min at 95-100°C and the supernatant collected for western blot analysis.

Salivary flow rate
To assess the salivary flow rate, NOD/ShiLtJ mice were weighed, and mild anesthesia was induced with a solution of ketamine (100 mg/mL; Beijing Double-Crane Pharmaceutical Co., China) and xylazine (20 mg/mL; Sigma, St. Louis, MO, USA) in sterile water, given intraperitoneally (1 μL/g body weight). Salivary secretion was stimulated using 0.1 mL/kg body weight of a pilocarpine solution (50 mg/mL; Beijing Double-Crane Pharmaceutical Co.) subcutaneously. Saliva collection began within 2 min of pilocarpine administration. Animals were positioned with a 75-mm hematocrit tube placed in the oral cavity, and whole saliva was collected into pre-weighed 0.75-mL tubes for 10 min. The amount of saliva collected was determined gravimetrically.

T-cell staining and flow cytometry
Intranuclear staining was carried out with Fixation/ Permeabilization buffer solution (eBioscience) according to the manufacturer's instructions. For intracellular cytokine staining, cells were stimulated with phorbol myristate acetate (10 ng/mL), ionomycin (250 ng/mL) and Golgi-Plug (1:1000 dilution; BD Biosciences, New Jersey, USA) at 37°C for 4 h, and then fixed with Fixation/Permeabilization buffer solution (BD Biosciences) according to the manufacturer's instructions. Stained cells were analyzed on a FACSCalibur or LSRFortessa (both from BD Biosciences), and data were analyzed with FlowJo software (BD Biosciences).

Total RNA extraction and microarray assay
BMMSCs from WT mice and Id3 −/− mice were seeded in 10-cm 2 dishes, cultured until they reached 80% confluence. Total RNA was extracted using TRIzol Reagent (Life Technologies, California, USA). Following purification with an RNeasy kit (Qiagen, Valencia, CA, USA), cDNA was generated using One-Cycle Target Labeling

Histological analysis of submandibular glands
Submandibular gland samples were fixed with 4% paraformaldehyde at 4°C for 24 h, embedded in paraffin, and sliced into 5-μm thick sections. After hematoxylin/ eosin (H&E) staining, sections were photographed at room temperature using a microscope (Olympus BX51, Japan), and the area of inflammatory foci (containing ≥50 lymphocytes per 4 mm 2 tissue) was calculated per field at ×200 magnification (×20 objective lens) using Image-Pro Plus software v6.0 (Media Cybernetics, Maryland, USA). Five entire salivary gland sections for each animal were counted, with an average of 10 fields/ gland, by an experienced expert in histopathology in a blinded fashion.

Statistics
All experiments are randomized into groups of similar sample size by block randomization. Data collection and analysis were performed blindly. No samples and animals were excluded from analysis. All experiments were performed at least in triplicates. Comparisons between two groups were performed using unpaired two-tailed Student's t tests; one-way analysis of variance (ANOVA; with Tukey's multiple comparison post hoc test, Post tests were run only if F achieved P < 0.05.) was used for comparisons between more than two groups. P values < 0.05 were considered statistically significant. Statistical analyses were carried out using GraphPad Prism 6 (California, USA). Statistical power analysis was used to ensure adequate sample size for detecting significant difference between samples. The variance is similar between groups that are being statistically compared.